The journey from data to feature

By James Whitlam - September 29, 2020

Solutions offered by InsurTech’s must be simple and intuitive without compensating on the layers of insight that drive them. InsurTech’s are ideal partners for insurers due to the time and skill required to create viable solutions. It can be easy to overlook how much groundwork is needed to build the most straightforward-looking feature in a digital platform. Here, I’ll outline the data sourcing process that leads to a feature, adding context with the example of viewing a vessel’s live location on a map.

The journey from data source to functioning feature follows four basic phases:

  1. The need
  2. Market evaluation
  3. Qualification and negotiation
  4. Interpretation

Each of these phases require collaboration with numerous stakeholders, both inside and outside of Concirrus. A phase can take weeks, or even months, to complete.


Phase one – The need

This is the most strategic and crucial stage; failure to identify the correct issue and prescribe the correct solution leads to a host of inefficiencies. To ensure exact needs are identified, a collaborative approach is adopted that includes internal teams, partners, and customers. Suitable data sources are often required to enable changes to the product or customer experience. When a need for development is identified, typical questions arise, such as:


What specific data will the data science team need to build a new and reliable predictive feature?

Should we prioritise the depth or the frequency of data to give the best possible insights?


In the case of vessel locations, we needed to understand:

  1. The kind of vessels that needed to be identified
  2. Where in the world would we need visibility
  3. How frequent the positions needed to be received

This process highlighted key hurdles that would need to be addressed in the following phases. Specifically, the need to collate the most comprehensive AIS data possible, and the need for a solution to the issue of vessel readings becoming less reliable in certain locations around the world, or in high traffic density areas.


Phase two – Market evaluation

This is one of the most challenging parts of the process. Once the required data is identified, it is not always straight-forward to deliver.

There is an impressive range of data available in the market, however specific requirements mean a solution rarely comes from a single source. The task, therefore, becomes one of isolating specific data sets from a wide range of sources. Relationships with multiple third parties allow us to discuss exactly which elements of their data is applicable, and what kind of agreement can be established around supplying that data.

With live vessel locations, a large portion of the world’s fleet transponder signals are sourced from terrestrial receivers. However, in order to overcome gaps due to geographical limitations or traffic dense regions, we needed to incorporate data from near-Earth orbit satellites. Spire, who operate a constellation of nanosatellites, provide this dataset, enabling us to build a complete picture of the world’s shipping movements.


Phase three – Qualification and negotiation

Once potential providers are identified, we verify the utility of sample data and understand exactly how wide reaching the data is. It’s probably the most time-consuming phase, as we screen, qualify and negotiate. We’ve historically ended up signing an agreement with two out of every 60 candidates we consider. It is an arduous task, but one that ensures that the data that underpins our products is both rich and highly accurate.


Phase four – Interpretation

Once the data is acquired, Data Strategy hand over to Data Science, who apply a ‘cleanse and enrichment’ process to ensure that raw data from multiple sources can be interpreted and displayed uniformly, without error, in-platform. Vessel locations were no exception, with feeds from terrestrial and satellites sources having to be combined to create a coherent stream of information.

With the data fully prepared, and any applicable modelling or analysis carried out, Data Science hand over to Product Development. This is where customer-facing tools are developed, presenting the data clearly in the form of a feature. With vessel locations in mind, data was displayed in a map view, including search and alert functions.


In the case of Quest Marine Hull, being able to see a vessel’s location on a map is the platform’s most basic function, but is the result of a multi-team endeavour requiring considerable expertise in various fields. It is a strong example of how something simple is derived from complexity. There are no shortcuts to creating this kind of functionality, but by partnering with a suitable InsurTech there can be a straightforward way to access it.

To learn more about the three reasons to collaborate with an InsurTech here


Contact Us



Powering the future of insurance

If you want to book a 1-1 meeting with us then let us know by clicking the link below.

Book a meeting